All you never wanted to know about 8 inch floppy drives

We connect an 8" floppy drive to a PC, learn the 11 ways they are different from standard PC floppies, and even design an adapter and make it available for all on Tindie.

Chapters:
00:00 Intro
01:20 My 8" drives come from an HP 9895
02:37 The IBM 8" floppy and its media
06:03 Antoine's 8" floppy adapter
07:06 Booting my PC from the 8" floppy
08:24 8" drive restoration and bring up
11:08 Making the prototype cable adapter
20:16 The 11 quirks of 8" floppies
20:52 Single sided vs. double sided
21:48 Write protect works reverse
22:52 Four address wires!
24:56 Head load signal
25:41 Door lock signal
26:38 Disk ready vs. disk change
28:02 Track 43 current control
28:37 Pull up resistors
29:28 500kHz data rate
30:07 FM modulation support
30:49 ImageDisk and Omnidisk setup for 77 tracks
32:52 Formatting an 8" Floppy
34:07 Using the floppy with Windows
35:25 Playing Donkey Kong from the 8" diskette

Stuff that supports the channel:

Our lovely sponsors

[Rob]

I have done a lot of tinkering with those IBM drives in the 1980's...
The drive you show at 2:48 is the actual drive that was in those controllers and computers of the day to load the microcode.
It was behind a door on the cabinet and was only supposed to be accessed by the Customer Engineer (a technician that usually was resident on site at important installations to quickly repair things).
Note that this type of drive has no front bezel, it just hinges open to insert/remove the diskette. There is nothing like a door lock, just a bulky manually operated catch.
Also no things like a disk activity LED. The "early drive" you show is actually from the later use of that diskette in data entry stations (where data that would normally be punched on cards was written on these floppies). Those were the first ones that had front bezels, LED, door locking etc.

The stepper motor is very slow, like 50-70ms track to track access time. It is linked to the head movement via a Malthezer Cross, a linkage that locks the heads in place when the stepper motor is not energized (it transfers rotation in the direction motor->heads but blocks rotation of the heads when the motor is not energized).
The step mechanism is VERY noisy! You would not hear it in a typical mainframe server room of course, but you certainly hear it when used at home (on the next floor, even). Compare to a teletype :-)
The drive had a proprietary IBM interface connector with signals similar to the later Shugart standard, but indeed with 150ohm pullups. I made interface boards that converted from that standard to the Shugart standard, which mainly consisted of beefy TTL drivers to drive the pullups.
In those days I had a TRS-80 model 1 and I used NEWDOS/80 2.0. It had sophisticated configuration of the drive parameters. You could arbitrarily set the number of cylinders, sides, sectors per track, steprate, and even "double step".
So my interface board included a divide-by-two on the "step" signal, I configured "double step" in the operating system and set it to the slowest steprate so the double step was slow enough for the drive to keep up.
I modified the floppy controller to have 500kHz mode (as opposed to the 250kHz used on 5.25" drives) and with all this I was able to use this drive in NEWDOS/80 as a standard drive, but with more capacity (the standard drives for that computer were 5.25" single sided single density 35 tracks!)
I still have a couple of them in storage, 3 in a big cabinet with powersupply and 50pin ribbon cable. I intend to power it up sometime and use GreaseWeasle to read what is on them.

[rainiergruber]

I love how we are complaining about faint coil noises in our graphics cards nowadays, and this thing simply sounds like a tractor is pulling up the driveway.

[UsagiElectric]

Excellent!
I love seeing more CDC floppies out and about. The 9406 is actually what I needed for the Centurion since it supports seek pulses as fast as 3ms, but the 9400 that I bought uses a different type of stepper mechanism and only supports seek pulses as fast as 10ms. We had to do a little microcode editing on the controller card to slow it down.
But, I've spent so much time with my head buried in the 9400, I've never actually seen what the 9406 looked like. It's really fascinating how much the CDC/Mag Peripherals design changed in such a short time!

[douro]

Magnetic Peripherals was later spun off from CDC, becoming Imprimis, and still being based in Oklahoma City. It eventually became part of Seagate who adopted CDC/Imprimis' practice of using the names of birds and predatory fish to name their storage products.

[Scrizati]

Between this and Adrian's video we have so much more information that is easily available now!

[vwestlife]

You can also specify 77 tracks with the DOS FORMAT command by using the /T:77 parameter.

[ignaciomenendez]

In IBM San José Engineering lab, we wrote the 8” diskettes for the 2835 microcode, in a modified 2841 File Control Unit, with the required floppy microcode on modified TROS tapes.
We had a special 8” diskette attached to the 2841 frame, that had R/W capability.
This is the way we tested microcode and sent patches for the 2835/2305 high speed drum facility (which was actually made up of many disk platters, with multiple heads for track on each platter, and no physical head movement.

[darrylr]

I spent far too much time shuffling 8" floppy drives in PDP-11s. Lots of Shugart floppy drives, they were built like a tank and very reliable at the time considering how we abused them.

[paulcohen]

There were two more types of floppies:
Hard sectored, which had a physical hole for EACH SECTOR in addition to the index.
Soft sectored, with only one hole, the index as mentioned in the movie.
(Personal knowledge, worked with both types).

[Satelitko]

"How should we name this amazing achievement in data storage technology? Well, it's a disk, and it's floppy... floppy disk?" I love how scientists and engineers name stuff :D

[nickm]

Haha - this brings back memories - I was a DEC service engineer in the mid 80's, some pdp11's I looked after had RX01/RX02 8 inch floppy drives and a few pdp8s did too. They made a lovely clunking noise! I seem to remember they used pre-formatted IBM disks, and I think they could be read on an MS-DOS system too.

[WillBreaksStuff]

Thank you! I was maybe 10 (1991) and my father took me to an auction at a local community college. There was a desk with a terminal built into it and I’ve had the shape and screen size burnt into my memory but could not for the life of me place it. I could never find any information about it. You showed an image of an IBM 3470 and that was it exactly! I even remember they had another system that was two bolted together that shared an 8 line display?

Anyhow. Thanks again. I finally know the name of that amazing looking terminal I’ve lusted over for over 30 years!

[ordinosaurs]

8" drives were already on the way out when I got interested in computers, but my first system had one attached (originally for data interchange between clients' older systems and the computer). You don't look a given hose... especially when said horse comes with a one cubic meter crate full of 1.2 MB 8" disks and your allowance isn't enough to cover a box of 5"1/4 disks per month. So for a couple of years, I used this drive a lot. Still works, btw.

[Dennis-ucgm]

This brings back memories of the late 80's for me . I spent many hours putting in circuit nodes for a Gen Rad bed of nails tester. We'd save all our work on 8 inch floppy's and then at some point we'd transfer the files to a DEC main frame via a 1200 baud modem. I remember when we needed a fresh floppy you'd have to see a "gate keeper" guy that would dole them out sparingly. By today's standard these we're overbuilt and made them very survivable, yes you'll still have to replace components that normally succumb to deterioration. Great video on a nostalgic subject. 👍

[MattTester]

Lovely to see the Dolch making an appearance again, mine has had a two year sleep but I should really wake it up.

[mysockC]

This takes me back! Those good old days where sometimes stuff worked, sometimes it didn’t, and sometimes you used the wrong interface and the whole thing went up in smoke. Still there was something special about using things like my first computer, an 8086 I built from dumpster parts with a 256 color 8-bit VGA card, (or 6809s and HC11s used in very early car ECUs in classic GMs), not to mention the family C64 as a kid and the Atari console my brother had, or the TRS80 that sometimes saw action. Much more interactive time where you still had to do more than just turn it on to use it.

[michaelhaardt]

To make things worse, there had been some drives which used a binary addressing scheme on the DS jumpers to allow up to 16 drives. Those resistor packs are termination resistors that should only be present on the last drive (floppy cables can be LONG), that's why they are socketed. Great episode and for the first time I did not learn something new, but had a pleasant summary of things that drove people nuts 40 years ago already.

[bazoo]

2:01 - A grey bearded nitpick here: HP-IB was standardized as IEEE-488 GP-IB which indeed saw quite widespread use, but mostly in instrumentation. A metrology lab I work with still uses it, and one of the more surprising uses was as peripheral bus for Commodore (then CBM) PET series of 6502-based microcomputers. One of such peripherals was their dual drive intelligent 5, 25" floppy unit (it contained two 65xx processors in the role of microcontrollers, and communicated with the host using a very high level ASCII protocol.)

Weird.

[TRA]

Back in the ‘80s, when working at HP, I wrote low-level code to recover data (written in an incompatible format) on 8” disks using a 9895. Good times 😊

[qzorn]

wow Our company had a lot of these large 8 inch floppy drives and they worked fairly well. The big thing was pre internet days to track down parts and information. Grrr. 😎 Thanks.